# Articles to 2015-05-21

First the link to this week’s complete list as HTML and as PDF.

Camps et al. give the total yearly supply of a photovoltaic generator in kW/h and the number of nominal peak-sunshine hours per year for Barcelona as 1.6 instead of 1600 (figures 9–11). Nagasawa et al. seem to consider a change of 100 % to be synonymous to “no change”. And Paolo et al. call an absolute thickness given from an arbitrary zero-point a “thickness change” in their figure 3. This is like calling a temperature given in degrees Celsius a “temperature change” or an elevation relative to mean sea level an “elevation change”. The figure caption corroborates the label by adding the missing time dimension in stating “Dots represent average thickness change every 3 months”. Only a very careful reading of the whole text can prevent us from coming to a completely wrong conclusion.

Commenting on last week’s equality study by Williams & Ceci, Bernstein notes a higher attrition rate. It is a known and well documented fact that those striving for the top accept an extremely strenuous and stressful life of eighty and more hours per week. Isn’t it just possible that instead of an outside bias women are sensible enough to decide against that kind of life, as incidentally do about 90 % of men in middle management?

Grigoryan strongly reminds me of lectures in business economics. You define a bunch of brain-dead oversimplified parameters that you can’t measure and are hard put to guess to an order of magnitude. You combine them in a simple equation of middle school level mathematics. You then solve this equation and pretend to have gained some new and deep understanding.

As always the important numbers are missing from Holmes, but they can be estimated. What we’re given are 800 000 deaths per year from colorectal cancer worldwide and a yearly incidence of 50 per 100 000, the highest in the world, for Slovakia and 20 per 100 000 deaths. With 8 milliard people and a life expectancy of 50 years we expect 160 million deaths per year of which 800 000 are 0.5 %. With a life expectancy of 80 years in Slovakia we would have 1250 deaths per 100 000 per year of which 20 are 1.6 %. Interestingly obesity raises the risk by only a third while exercise lowers it by a fifth, which are both lower than the general health effects. So fat couch potatoes run a lower risk of it being CRC to get them than the general population. Coloscopy being a million dollar enterprise needs to advertise and it seems that you need to take those advertisements with a pinch of salt just like all others.

Hall et al. finally supply the primary source for the comment by Servick (list of 2014-10-11). It seems they have found the genetic base for susceptibility to the placebo effect. Does this simplify clinical trials or just add more confounders?

Southern California seems to have a pronounced 24 h cyclicality for wind power supply that is furthermore exactly in phase with grid demand, at least for the seven days Safdarnejad et al. have chosen to base their study on. These are very special conditions that obtain nearly nowhere else in the world, so whatever the merits of the study itself may be, it has to be completely irrelevant to just about everyone.
Secondly we see here another politically correct attempt at the perfect solution. In all technology there is the very strong law of diminishing returns. So it’s an utter nonsense to equip one power station for 90 % capture when much less effort could fit two for 50 %. After you have achieved saturation over all the plants and gained a lot of experience on the way, then is the time to raise the retention rate.

The most important result in Pearson et al. seems to be, that human food, both plants and in this case animals, comes in a wide range of isotopic values. Thus elucidating diet from isotope values is far less straightforward than it initially seemed years ago when the method was young.

No society in the history of mankind has ever been able to stave off hunger and provide safety against the vagaries of the climate – they only thought they had due to an exceptionally long stretch of clement weather. Munoz et al. provide another example, this one for well over 500 years. There is no reason to assume our current civilisation to be any different in the long term.

Liu, Schunn et al. display my standard beef with psychological studies. If there was a single, standardized, precise norm-human whom you can only describe with imprecise measurements, then there exists a single true value and you can raise your precision by making more measurements. If that were the case, then the use of the standard error as employed here would be appropriate. If however you have a whole bunch of individual humans, each reacting individually and differently to the test applied, then a larger number of probands will capture more of the range present in the total population. In this case, which I take to be the more appropriate representation, only the standard deviation will adequately represent the effect on the group as a whole. An artificially small SE will raise expectations for a result for every single pupil that will not materialize in practice. It will make teachers feel bad through no fault of their own.